View source for Talk:Jetson/Installing OpenCV
|Thread title||Replies||Last modified|
|Is anyone modifying / update this wiki?||2||16:34, 20 August 2016|
|how to avoid conflicts?||1||21:05, 30 September 2014|
|problems with OpenCV install on Jetson TK!. (CUDA_CUDA_LIBRARY undefined)||1||17:41, 18 September 2014|
It seems there are code examples on this wiki that are no longer applicable or are out of date with current versions. Who updates this page with the latest info?
In looking at page history you can see that updates to the page were made as recently as July
Yes I wrote nearly the entire Jetson TK1 Wiki and I still update it when I find issues, but I don't work for NVIDIA anymore and I'm not using Jetson TK1 much at the moment, so I don't update it often. I actually used these OpenCV install instructions on Jetson TK1 just a few weeks ago, so it certainly is up-to-date, even for OpenCV 3.0, but of course it might not be perfect, there are a lot of combinations of L4T and OpenCV I haven't tested.
Jetson TK1 doesn't normally come with OpenCV, so this page tries to describe the different options you have for putting OpenCV on your Jetson TK1. And at the bottom it links to the OpenCV Tutorial that shows how to get started actually using OpenCV, after you've installed it on your Jetson TK1.
If you've tried following this Wiki page and are having trouble, the best place to ask is NVIDIA's Jetson TK1 forum at "http://devtalk.nvidia.com/default/board/139/embedded-systems/".
You do not have permission to edit this page, for the following reasons:
- The action you have requested is limited to users in the group: Users.
- You must confirm your email address before editing pages. Please set and validate your email address through your user preferences.
You can view and copy the source of this page.
CUDA itself is functional, OpenCV-2.4.9.zip download as described here. But ccmake reports CUDA_CUDA_LIBRARY undefined, numpy not present, etc. (In default tegraK1 ubuntu out-of-box)
It sounds like you are trying to build OpenCV from source? If you do that, you can't get the hundreds of CPU optimizations (for quad-core and NEON SIMD) that OpenCV4Tegra prebuilt packages provide. So I recommend that you don't build OpenCV from source, I recommend you use the OpenCV4Tegra prebuilt packages, as mentioned above the manual installation section. OpenCV manual installation is mostly just for people modifying the OpenCV library (such as NVIDIA employees, since we have access to the OpenCV4Tegra source code including all our closed-source optimizations).